Skip to main content

Decentralized data processing for IoT platforms

Project description


Joule is a data capture and signal processing engine. It allows you to build robust sensors using embedded systems such as the Raspberry Pi. Joule uses modules to build complex acquisition and signal processing workflows from simple building blocks. Modules are user defined processes that are connected together by data streams.

Joule acts as a process manager, ensuring that modules start at system boot and are restarted if they fail. Joule also collects runtime statistics and logs for each module making it easy to detect bugs and find bottlenecks in processing pipelines.

See for full documentation. To get started see:


Streams are timestamped data flows that connect modules together. Streams can represent primary measurements such as readings from a current sensor or derived measurements such as harmonic content. A stream has one or more elements and can be viewed as a database table:

========= ====== ====== === ======
timestamp value1 value2 ... valueN
========= ====== ====== === ======
1003421   0.0    10.5   ... 2.3
1003423   1.0    -8.0   ... 2.3
1003429   8.0    12.5   ... 2.3
1003485   4.0    83.5   ... 2.3
========= ====== ====== === ======


Modules process streams. A module may receive zero, one or more input streams and may produce zero, one, or more output streams. While Joule does not enforce any structure on modules, we suggest structuring your data pipeline with two types of modules: Readers, and Filters. Readers take no inputs. They directly manage a sensor (eg a TTY USB device) and generate an output data stream with sensor values. Filters take these streams as inputs and produce new outputs. Filters can be chained to produce complex behavior from simple, reusable building blocks.


Using a light sensor and temperature sensor to detect occupancy in a room:

[Module]  a_stream

[ TempReader] --> temp_values  --,
[LightReader] --> light_values --+--> [OccupancyFilter] --> room_status


  • Fast, lightweight

  • Highbandwidth signal processing


Joule requires Python 3.10 or later. Install Joule by running:

$> pip install -r requirements.txt $> pip install .

It is also available through PyPi:

$> python3 -m pip install joule

To run the Joule daemon, you must have a PostgreSQL database running with the TimescaleDB extension installed. Full instructions on installing TimescaleDB can be found at A script to automate the installation process on Ubuntu is included in this repository:

$> sudo ./

A script to create a database with the TimescaleDB extension is also included:

$> sudo -u postgres psql < ./postgresql-bootstrap.sql

NOTE: It is strongly recommended to change the default password in the bootstrap script before running it. Once the database has been configured you can initialize Joule using the appropriate DSN string:

$> sudo joule admin initialize –dsn joule:joule@localhost:5432/joule

Change the password to match your configuration. Finally, use systemctl to manage the daemon service:

$> sudo systemctl [enable|disable|start|stop|status] joule

To connect to the Joule daemon, use the command line client:

$> sudo -E joule admin authorize

See for full documentation on using Joule.


To run unittests and collect coverage information:

$> coverage run -m unittest $> coverage html $> python3 -m http.server –directory htmlcov

To run integration tests you must have Docker and access to the nilmdb image (for nilmdb tests)

$ joule/tests/e2e> ./ [timescale|nilmdb]

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

joule-0.9.49.tar.gz (6.5 MB view hashes)

Uploaded source

Built Distribution

joule-0.9.49-py3-none-any.whl (545.6 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page